Impact Evaluation Associated with the Bangladesh Agricultural Value Chains project

Last registered on June 15, 2016

View Trial History

Pre-Trial

Trial Information

General Information

Title

Impact Evaluation Associated with the Bangladesh Agricultural Value Chains project

RCT ID

AEARCTR-0001295

Initial registration date

June 15, 2016

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published

June 15, 2016, 1:23 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Country

Bangladesh

Region

Khulna, Barisal

Primary Investigator

Name

Alan de Brauw

Affiliation

International Food Policy Research Institute

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Hazel Malapit

PI Affiliation

International Food Policy Research Institute

Contact Investigator

PI Name

Berber Kramer

PI Affiliation

International Food Policy Research Institute

Contact Investigator

Additional Trial Information

Status

On going

Start date

2016-01-04

End date

2018-03-30

Keywords

Agriculture

Additional Keywords

value chains, gender

JEL code(s)

O12, O13, Q12, J16

Secondary IDs

Abstract

In this impact evaluation, we are partnering with the Bangladesh Agricultural Value Chains Project to test interventions within specific value chains for impacts on outcomes among smallholders. The goal of the interventions being studied is to move the value chain from an equilibrium in which farmers both face uncertainty about input quality and crop marketing methods to an equilibrium in which high quality inputs are available and used by smallholders, and they are better able to receive high prices for their output as it is of less variable quality. For the jute value chain, the impact evaluation centers around cluster randomizing both training and promotional events for high quality, certified, and branded fertilizer. The promotional events included a raffle for discounts for fertilizer; main outcomes to be studied include whether or not treated households participated in the trainings or promotional events; use of improved inputs; attitudes about improved inputs and trust in input retailers or input brands; specific crop production and yields; and agricultural revenue and profits.

External Link(s)

Registration Citation

Citation

de Brauw, Alan, Berber Kramer and Hazel Malapit. 2016. "Impact Evaluation Associated with the Bangladesh Agricultural Value Chains project." AEA RCT Registry. June 15. https://doi.org/10.1257/rct.1295-1.0

Former Citation

Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Experimental Details

Interventions

Intervention(s)

We are working with the Bangladesh Agricultural Value Chains project to pilot test randomized interventions to observe their impacts on productivity of specific crops- jute and mung beans. In the jute value chain, we are specifically studying the impacts of trainings conducted by NGOs and of promotional activities conducted by input dealers to market improved or reliable inputs to smallholders. The interventions are being implemented as a cluster randomized control trial; the two interventions described above were cross-randomized. Mung bean interventions are in the planning stage for the 2017 growing season. More complete descriptions of the interventions are available from the principal investigators.

Intervention Start Date

2016-03-07

Intervention End Date

2017-10-02

Primary Outcomes

Primary Outcomes (end points)

Key outcomes include participation; input purchases, input use, agricultural productivity, consumption expenditures, and food security.

Primary Outcomes (explanation)

We will use the Women's Empowerment in Agriculture index as an additional outcome; its construction is discussed at the WEAI Resource Center (https://www.ifpri.org/topic/weai-resource-center).

Secondary Outcomes

Secondary Outcomes (end points)

Secondary Outcomes (explanation)

Experimental Design

We will have similar experimental designs embedded in two value chains, those for jute and mung beans. Both are part of the larger AVC project. In the jute value chain, For jute, villages were first randomized into two groups: 25 villages that will receive training and 25 that will not. Villages were then additionally randomized into promotion and non-promotion groups, yielding the following four treatment groups:

1. Training + Promotion (13 villages)
2. Training + No Promotion (12 villages)
3. No Training + Promotion (13 villages)
4. No Training + No Promotion (12 villages)

Within the promotion villages, a raffle was held for households to receive coupons towards the purchase of specifically branded NPKS fertilizer which is helpful for jute cultivation specifically. The discounts were 80%, 50%, and 20%. The raffle was specifically held among farmers in the promotional groups. Therefore we can calculate statistical power in terms of either participation in one arm of the cluster RCT or in terms of categories of lottery winners.

Experimental Design Details

Randomization Method

Randomization was done by computer and is still to be conducted for the mung bean groups.

Randomization Unit

Village

Was the treatment clustered?

Yes

Experiment Characteristics

Sample size: planned number of clusters

50 clusters for jute value chain, 50 clusters for the mung bean value chain

Sample size: planned number of observations

2000 household level observations (both males and females were interviewed in households when available, so the targeted sample size was actually 4000 interviews)

Sample size (or number of clusters) by treatment arms

See above.

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

For the jute intervention, we compute the minimum detectable effect for discrete variables generally as a function of the baseline adoption rate. For smaller initial rates, the minimum detectable effect is smaller in terms of percentage points; for example, if there is 10 percent adoption, an 11 percentage point difference is detectable. If the initial adoption rate is higher, a larger percentage point difference is needed to identify outcomes, but the percent change need not be as large. For jute yields, we can identify a 13 percent difference from the control group in any specific arm of the intervention. Among raffle winners, the difference is 23 percent; among 3rd place winners, 16 percent. We are more likely to be able to detect the impacts of winning the raffle or coming in 2nd place because they are more likely to redeem coupons. In general, for a continuous variable the MDE is 0.31 standard deviations.

Supporting Documents and Materials

IRB